Politics | Christian democracy » Golden-Chang - Competitive Corruption, Factional Conflict and Political Malfeasance in Postwar Italian Christian Democracy

Datasheet

Year, pagecount:2001, 35 page(s)

Language:English

Downloads:4

Uploaded:April 30, 2018

Size:1 MB

Institution:
-

Comments:

Attachment:-

Download in PDF:Please log in!



Comments

No comments yet. You can be the first!

Content extract

v53.i4588golden 9/27/01 5:20 PM Page 588 Source: http://www.doksinet COMPETITIVE CORRUPTION Factional Conflict and Political Malfeasance in Postwar Italian Christian Democracy By MIRIAM A. GOLDEN and ERIC C C CHANG* T HIS article is motivated by three central theoretical puzzles regarding systemic political corruption in democratic settings. First, political corruption may involve high-level politicians and in extreme cases may involve even an entire political class. As such, it is clearly not a problem of inadequate or ineffective monitoring by politicians of the bureaucrats they supervise, as is the case with the more commonly studied phenomenon of bureaucratic corruption.l Second, systemic political corruption in a democratic context occurs despite a wealth of legal regulations intended to prevent it. The cases of persistent, highlevel corruption of which we are awarerecent revelations feature Germany and France, as well as Italydo not occur in settings that exhibit

substantially less transparency or other judicial or constitutional characteristics that obviously account for their unusual outcomes. Finally, Schumpeterian competition between political elites ought to prevent large-scale, persistent political corruption from taking root in democratic polities in the first place, precisely because honest partisan competitors can always offer their services to voters and ought to be preferred over their dishonest counterparts. The very existence of wide* Earlier versions of this paper were delivered at the annual meetings of the American Political Science Association, Atlanta, September 2–5, 1999, at a meeting of the MacArthur Research Network on Inequality and Economic Performance, MIT, October 1–3, 1999, and at the Workshop on the Political Economy of Europe, Center for European and Russian Studies, UCLA, February 3, 2000. Data analyzed in this study were generously provided by Franco Cazzola and by the Gruppo Democratici di Sinistra-L’Ulivo

of the Italian Senate. We are grateful to Gianfranco Pasquino for much useful advice and to John Agnew, Barry Ames, Kathleen Bawn, Gary Cox, Donatella della Porta, Scott Desposato, Carlo Guarnieri, Richard S. Katz, John Londregan, Carol Mershon, Ronald Rogowski, Jean-Laurent Rosenthal, Daniel Treisman, George Tsebelis, Alberto Vannucci, Carolyn Wong, Alan Zuckerman, and other members of UCLA’s Tuesday Political Economy lunch group for comments on earlier versions. For assistance, we thank Judit Bartha, Michael Lin, and David Yamanishi For financial support, Golden acknowledges the National Science Foundation (SES-0074860) and the Academic Senate of the University of California at Los Angeles, as well as the hospitality of the Russell Sage Foundation, New York. The authors are solely responsible for the work reported here 1 For the distinction between political and bureaucratic corruption, see Susan Rose-Ackerman, Corruption: A Study in Political Economy (New York: Academic Press,

1978). World Politics 53 ( July 2001), 588–622 v53.i4588golden 9/27/01 5:20 PM Page 589 Source: http://www.doksinet COMPETITIVE CORRUP TION 589 spread, persistent political corruption in a competitive electoral system constitutes a puzzle. This leads us to ask when and why political competition may fail to inhibit rent seeking on the part of elected officialsspecifically, the illegal receipt of monies in the form of bribes and kickbacks that constitutes the core of political corruption.2 Persson, Tabellini, and Trebbi hypothesize that although electoral competition does not eliminate rent seeking, some types of electoral rules make elected officials more accountable than others and hence less prone to corrupt behaviors.3 They contend that party lists lead to more corruption than occurs in political systems in which individuals compete against each other for public office, because electoral systems characterized by the latter procedure allow voters to control individual

representatives. Others subscribe to same view.4 Students of countries that actually use proportional representation and party lists tend to believe the reverse, however. According to these scholars, electoral systems that provide incentives for politicians to develop personal reputations generate less accountability, higher rents, and potentially more corruption than those encouraging the development of party reputations, perhaps because parties wield instruments that can discipline their members.5 This view stands in direct contrast to the arguments advanced by Persson, Tabellini, and Trebbi. Our study contributes to this debate by systematically testing the hypothesis that competition among individual candidates from the same party is significantly associated with political corruption. While this hypothesis arises naturally from the theory advanced by Persson, 2 Although political corruption encompasses a broader set of activities, the receipt of funds in exchange for favors is

especially important, both because of the potentially distortionary economic effects of these activities and because they are almost always illegal. When elected officials persistently engage in illegal behavior, troubling consequences arise for the democratic order. 3 Torsten Persson, Guido Tabellini, and Francesco Trebbi, “Electoral Rules and Corruption” (Manuscript, Institute for International Economic Studies, Stockholm University, November 2000). 4 For instance, Paul Mitchell, “Voters and Their Representatives: Electoral Institutions and Delegation in Parliamentary Democracies,” European Journal of Political Research 37 (May 2000). 5 Those whose work supports this view include Barry Ames, “Electoral Strategy under Open-List Proportional Representation,” American Journal of Political Science 39 (May 1995); Bruce Cain, John Ferejohn, and Morris Fiorina, The Personal Vote: Constituency Service and Electoral Independence (Cambridge: Harvard University Press, 1987); John M.

Carey and Matthew Soberg Shugart, “Incentives to Cultivate a Personal Vote: A Rank Ordering of Electoral Formulas,” Electoral Studies 14 (December 1995); Morris P. Fiorina and Roger Noll, “Voters, Bureaucrats and Legislators: A Rational Choice Perspective on the Growth of Bureaucracy,” Journal of Public Economics 9 (April 1978); idem, “Voters, Legislators and Bureaucracy: Institutional Design in the Public Sector,” American Economic Review Papers and Proceedings 68 (May 1978); Barbara Geddes and Artur Ribeiro Neto, “Institutional Sources of Corruption in Brazil,” Third World Quarterly 13, no. 4 (1992); and David J Samuels, “Incentives to Cultivate a Party Vote in Candidate-Centric Electoral Systems: Evidence from Brazil,” Comparative Political Studies 32 ( June 1999). v53.i4588golden 9/27/01 5:20 PM Page 590 Source: http://www.doksinet 590 WORLD POLITICS Tabellini, and Trebbi, they do not test it. Instead, they compare the extent of reported political

corruption in countries using majoritarian electoral systems with its extent in those using proportional representation (PR). Theories of what Cain, Ferejohn, and Fiorina have called “the personal vote”6 suggest however that the more useful distinction lies between systems in which parties control candidate selectionclosed-list PRand those in which they do not, including majoritarian systems with primary elections and other two-ballot systems, single nontransferable vote (SNTV) systems, and open-list PR. This theory classes together electoral systems using SNTV (such as Japan), those using open-list PR (such as Italy), and those using two-ballot systems in which candidates of the same party compete against each other on the first ballot (such as France and the United States). The “personal vote” literature typically focuses on how politicians provide constituency service in order to gain individual recognition from voters, although the circumstances under which constituency

service also involves patronage and, in the extreme, outright corruption have not been identified. Plausibly, however, incomplete or blocked partisan competition for public office is a contributing factor, along with a legal environment that easily criminalizes campaign fund-raising efforts or contributions.7 Cross-national testing of these arguments requires building an appropriate data seta difficult task, given the complexity of ranking electoral systems according to their degree of campaign and candidate personalism.8 In the meantime, we draw on data from Italy’s postwar legislative elections, specifically examining the association between charges of corruption and intraparty competition in the country’s major governing party, Italian Christian Democracy (DC), during the postwar era. While there are some disadvantages to studying the impact of electoral rules on political corruption within a single national setting in which electoral rules themselves do not undergo

modification, these limitations are fewer than may initially seem to be the case.9 We take advantage of the rules in operation in Italy from after World War II until the 1994 elections (when the electoral system was modified) to investigate the impact of intraparty competitionor the search for votes by individual candidates within the DCon charges of malfeasance 6 Cain, Ferejohn, and Fiorina (fn. 5) See Miriam A. Golden, “The Effects of the Personal Vote on Political Patronage, Bureaucracy and Legislation in Postwar Italy” (Manuscript, University of California at Los Angeles, May 2001). 8 See Carey and Shugart (fn. 5) 9 Note as well that electoral systems are so rarely modified that studying the effects of change across time within countries generates only a handful of cases. 7 v53.i4588golden 9/27/01 5:20 PM Page 591 Source: http://www.doksinet COMPETITIVE CORRUP TION 591 against DC members of parliament. To anticipate our findings, we show that political corruption is

significantly associated with increases in intraparty competition. Although the determinants of political corruption are many, the present study, along with that of Persson, Tabellini, and Trebbi,10 are the only existing systematic analyses we know of that examine the impact of electoral institutions on political corruption. The Italian case was selected for four main reasons. First, Italy represents an extreme case of systemic political corruption in an established, democratic setting Rich (and therefore democratic) countries tend to enjoy relatively low levels of corruption, at least if surveys of international executives such as those conducted by Transparency International (TI) are to be believed.11 Per capita gross domestic product does a good job explaining perceived corruption among a sample of some fifty-odd countries analyzed by Treisman, for example.12 Italy, as Figure 1 shows, remains the most pronounced exception. It is the country for which wealth is least effective as the

single explanatory variable of perceived levels of corruption, as revealed by the fact that it is farthest off the regression line depicting a linear relationship between per capita gross domestic product and perceived levels of corruption. In fact, the studentized residual of Italy in the regression shown in Figure 1 is 3.1543, substantially above the normal threshold of 2 If we set Italy as a case indicator to distinguish it from other countries, the Italian coefficient in the expanded regression model is significantly different from 0 at less than the 0.01 level These results suggest that Italy is even exceptional among outliers. In the mid-1990s Italian corruption levels were far higher than the country’s relatively wealthy status would have predicted. Working with such an extreme case makes what is usually almost impossible to study a feasible area of empirical investigation Not only are the traits and mechanisms of corruption more visible in such a setting, but there is much

more of it to observe. Second, an implication of the finding reported above regarding Italy’s position in Figure 1that the country is so far off the regression lineis that unusual, nonrandom influences may be at work. While it 10 See fn. 3 The TI index fails to distinguish political from bureaucratic corruption, and the latter is undoubtedly more common than the former. We have no reason to believe that Italy’s location in the index would change, however, even if the index were confined to political corruption for the period through the mid-l990s; if anything, the Italian case would probably become relatively even more extreme. After the mid-1990s the TI index for Italy is probably capturing bureaucratic corruption, earlier political corruption, or some combination. See below (fn 87) for further discussion 12 Daniel Treisman, “The Causes of Corruption: A Cross-National Study,” Journal of Public Economics 76 ( June 2000); see also Alberto Ades and Rafael Di Tello, “Rents,

Competition, and Corruption,” American Economic Review 89 (September 1999). 11 v53.i4588golden 9/27/01 5:20 PM Page 592 Source: http://www.doksinet 592 WORLD POLITICS 10 TI Corruption Index 8 Italy 6 4 2 0 2.5 3 3.5 Log of GDP per Capita 4 4.5 FIGURE 1 THE RELATIONSHIP BETWEEN WEALTH AND PERCEIVED CORRUPTION IN VARIOUS COUNTRIES, C. 1990 SOURCE: Treisman (fn. 12) a The Transparency International index has been inverted so that higher scores indicate greater perceived corruption. would therefore be unwise to construct a general theory of political corruption in democratic settings based exclusively or even predominantly on information gleaned from the Italian case, any such theory would have to reckon with this caseeven if only to explain why it is so deviant. Anything we can reliably learn about the causes of corruption in postwar Italy will therefore ultimately contribute to theory building. Third, Italy offers a unique and heretofore underexploited source of data

on political corruption: requests by magistrates for authorization to proceed with an investigation against a member of parliament, or what are abbreviated in Italian as RAP (le richieste di autorizzazione a procedere). In essence, these are requests to suspend the parliamentary immunity of a legislator. These requests are made by investigating magistrates seeking to proceed with legal charges against an MP (whether a senator or a member of the Chamber of Deputies). Between 1948 and v53.i4588golden 9/27/01 5:20 PM Page 593 Source: http://www.doksinet COMPETITIVE CORRUP TION 593 1993, the Italian judiciary made 4,770 requests to parliament for authorization to proceed with such charges.13 Finally, the dramatic revelations of extremely widespread corruption among the Italian political class with the “Clean Hands” investigations that began in 1992 generated a rich literature describing the processes of Italian political corruption. While it was no secret even prior to 1992

that political parties in Italy had been engaged in extensive corruption (as a handful of studies had documented14 but as was also well known even in the absence of systematic research), after 1992 numerous aspects of Italian political corruption were scrutinized, generating a rich and detailed literature in both Italian15 and English.16 In addition, the revelations of corruption generated firsthand testimonials by some of the most knowledgeable participants.17 As a result, Italy offers the single most important source of information about the nature, extent, and workings of widespread political corruption in a modern democratic context. Despite the wealth of available literature, little is known with confidence about the causes of political corruptionin Italy or elsewhere. 13 Franco Cazzola and Massimo Morisi, “Magistratura e classe politica: due punti di osservazione specifici per una ricerca empirica,” Sociologia del Diritto 22, no. 1 (1995), 101–2 14 Most importantly, Franco

Cazzola, Della corruzione: fisiologia e patologia di un sistema politico (Bologna: Il Mulino, 1988); and Donatella della Porta, Lo scambio occulto: casi di corruzione politica in Italia (Bologna: Il Mulino, 1992). 15 Among others, Franco Cazzola, L’Italia del pizzo: fenomenologia della tangente quotidiana (Turin: Einaudi, 1992); Alessandro Pizzorno, “Introduzione: la corruzione nel sistema politico,” in della Porta (fn. 14); Franco Cazzola, “Storia e anatomia della corruzione in Italia,” Il Politico 13 (April 1993); Luca Ricolfi, L’ultimo Parlamento: sulla fine della prima Repubblica (Rome: La Nuova Italia Scientifica, 1993); Donatella della Porta and Alberto Vannucci, Corruzione politica e amministrazione pubblica: risorse, meccanismi, attori (Bologna: Il Mulino, 1994); Mauro Magatti, Corruzione politica e società italiana (Bologna: Il Mulino, 1996); and Alberto Vannucci, Il mercato della corruzione: i meccanismi dello scambio occulto in Italia (Milan: Società Aperta,

1997). 16 Including Donatella della Porta and Alberto Vannucci, “Politics, the Mafia, and the Market for Corrupt Exchange,” in Carol Mershon and Gianfranco Pasquino, eds., Italian Politics: Ending the First Republic, vol. 9 of Italian Politics: A Review (Boulder, Colo: Westview Press, 1995); Sarah Waters, “‘Tangentopoli’ and the Emergence of a New Political Order in Italy,” West European Politics 17 ( January 1994); David Moss, “Patronage Revisited: The Dynamics of Information and Reputation,” Journal of Modern Italian Studies 1 (Fall 1995); Donatella della Porta and Alessandro Pizzorno, “The Business Politicians: Reflections from a Study of Political Corruption,” Journal of Law and Society 23 (March 1996); David Hine, “Political Corruption in Italy,” in Walter Little and Eduardo PosadaCarbó, eds., Political Corruption in Europe and Latin America (Bassingstoke: Macmillan, 1996); Martin Rhodes, “Financing Political Parties in Italy: A Case of Systemic

Corruption,” West European Politics 20 ( January 1997); Stanton H. Burnett and Luca Mantovani, The Italian Guillotine: Operation Clean Hands and the Overthrow of Italy’s First Republic (Lanham, Md.: Rowman and Littlefield, 1998); and Donatella della Porta and Alberto Vannucci, Corrupt Exchanges: Actors, Resources, and Mechanisms of Political Corruption (New York: Aldine de Gruyter, 1999). 17 Gherardo Colombo, Il vizio della memoria (Milan: Feltrinelli, 1996); Piercamillo Davigo, La giubba del re. Intervista sulla corruzione, ed Davide Pinardi (Rome: Laterza, 1998); and Antonio Di Pietro, Intervista su Tangentopoli, interview with Giovanni Valentini (Rome: Laterza, 2000). v53.i4588golden 9/27/01 5:20 PM Page 594 Source: http://www.doksinet 594 WORLD POLITICS The Italian literature has focused mainly on systematizing the mechanisms and practices of widespread political corruption and on identifying and classifying the types of behaviors and actors involved in corrupt

exchanges. This has generated an unusually rich literature, often based on exhaustive primary research, detailing the practices of political corruption in postwar Italy. However, systematic hypothesis testing of the causes of widespread political corruption has been uncommon. The literature on developing countries, although older and often extensive, tended to be relatively unsystematic in its forms of analysis. Only recently have scholars begun to produce systematic work on the causes of corruption.18 This scholarship has been cross-national in scope, and hence the hypotheses examined have necessarily been limited by data availability. Thus, for example, the absence of data over time has precluded examining temporal change. This article, constrained by the limitations in the data that are available, offers a partial but systematic causal account of political corruption over the course of the postwar era in Italypartial in the sense that it explicitly tests the impact only of political

sources of systemic corruption. (Other possible types of causes, broadly categorized, include economic and cultural) To compensate to some degree, we supplement our causal investigation with an exploration of the temporal geography of political corruption in postwar Italy that allows a preliminary assessment of the prima facie plausibility of a rival cultural perspective. Despite current data limitation, our analysis generates some robust findings. We demonstrate that the unusually high levels of corruption observed in postwar Italy were in part an outgrowth of specific features of the open-list proportional representation electoral system in use there, features that promoted conflict between candidates from the same party. To document this, we assemble a database measuring intraparty competition in the country’s dominant political party, Christian Democracy. We analyze this in conjunction with a data set that we assembled on charges of malfeasance against DC members of parliament

The analysis shows that beginning in the early 1970s, a significant statistical relationship existed between intraparty conflict in the ruling Christian Democratic Party and the extent of suspected malfeasance among DC members of Italy’s Chamber of Deputies. Our analysis also finds that electoral competition with other parties shows no significant statistical impact on alleged malfeasance by DC legislators. We then examine the 18 Especially Ades and Di Tello (fn. 12); Persson, Tabellini, and Trebbi (fn 3); and Treisman (fn 12) v53.i4588golden 9/27/01 5:20 PM Page 595 Source: http://www.doksinet COMPETITIVE CORRUP TION 595 geography of political corruption over time in Italy, in order to assess whether corruption spread via a contagion process or whether it arose out of characteristics of the political system. The findings fail to corroborate a geographic or ancillary cultural perspective, again leading us to endorse an institutional interpretation. We offer evidence that

political corruption among Christian Democratic legislators appears to have become systematic about 1974, just when a law regarding the public financing of political parties was adopted. This suggests that legislation designed to limit discretionary financing of political parties may well have amplified the degree of corruption arising out of the electoral system. Corruption is operationalized in our analysis as alleged criminal wrongdoing. Although actual corruption, usually defined as the use of public office for personal or partisan gain, was undoubtedly more common than the proxy measure of suspected malfeasance employed in this study, data on actual corruption are, of course, not available. I. INSTITUTIONAL ORIGINS OF SYSTEMIC POLITICAL CORRUPTION We examine three hypotheses in this section. The first is that political corruption is in part an outgrowth of intraparty competition, which in turn is partially a function of specific characteristics of the electoral system. The second

is that interparty competition may also significantly affect political corruption. The third, which derives from our results on the first two, is that a legal change in the mid-1970s over the public financing of Italian parties may have magnified political corruption. We now detail the basic logic of these hypotheses. Systems of proportional representation, in which multiple representatives are elected from each district in proportion to the share of votes cast for their various parties, may use closed or open lists for ordering candidates. An open-list system forces candidates of the same party to compete against each other, because even though districts are multimember, the actual persons elected are decided not by the order established on the party list (which effectively allows parties to discipline their own candidates) but by the number of votes received by individual candidates. Katz has plausibly contended that this procedure pits party members against each other,19 creating

powerful incentives toward internal party disunity. A similar syndrome is associated with systems (such as that used in Japan until 1994) of the single, nontransferable 19 Richard S. Katz, “Intraparty Preference Voting,” in Bernard Grofman and Arend Lijphart, eds, Electoral Laws and Their Political Consequences (New York: Agathon Press, 1986). v53.i4588golden 9/27/01 5:20 PM Page 596 Source: http://www.doksinet 596 WORLD POLITICS vote. Both SNTV and open-list PR typically (but not inevitably)20 result in significant factionalism within parties. And factions, of course, need funds to compete effectively against one another. Typically, however, national party organizations have little interest in funding specific factions and instead direct resources mainly toward political competition with other parties. In Italy, for instance, public funding for political campaigns, established in 1974, directed funds to national party organizations, not to individual candidates.21 This

essentially leaves individual candidates and factions on their own to raise the necessary campaign funds. As Cox and Thies show using Japanese data,22 intraparty electoral competition is expensive, so incentives are strong to raise money. 23 Illicit campaign contributions are one result Anecdotal evidence suggests that such an interpretation of the electoral system may be useful for understanding the unusually high level of political corruption in postwar Italy. While it is impossible to quantify the extent to which political graft was used for intraparty competition, there is evidence that it was apparently extensive.24 One observer noted in the mid-1970s that “few Christian Democrats win seats without running personalized campaigns that are both extensive and costly.”25 Whole political careers were made on the basis of illicit monies. Della Porta and Vannucci, in their detailed study of the results of judicial investigations into corruption in Italy, argue that illicit

contributions to the Christian Democratsfor there were also many to other parties, especially the Italian Socialist Party (PSI)were aimed not at the DC as a whole but at particular factions within it.26 This suggests that it is worth making a more systematic investigation of the relationship between intraparty competition and political corruption. At the same time that intraparty competition may engender political 20 Samuels (fn. 5) Linda Bull, “Public Money, Political Parties, and Corruption: The Italian Case,” Italian Politics and Society: The Review of the Conference Group on Italian Politics and Society 48 (Autumn 1997). The putative impact of this law is analyzed below. 22 Gary W. Cox and Michael F Thies, “The Cost of Intraparty Competition: The Single, NonTransferable Vote and Money Politics in Japan,” Comparative Political Studies 31 ( June 1998) 23 Documented in Eric C. C Chang, “The Institutional Roots of Political Corruption in Postwar Japan” (Manuscript,

Department of Political Science, University of California at Los Angeles, January 2000); and Gary W. Cox and Michael F Thies, “How Much Does Money Matter: ‘Buying’ Votes in Japan, 1967–1990,” Comparative Political Studies 33 (February 2000). 24 Della Porta and Vannucci (fn. 15), 225–26 25 Donald Wertman, “The Italian Electoral Process: The Elections of June 1976,” in Howard R. Penniman, ed., Italy at the Polls: The Parliamentary Elections of 1976 (Washington, DC: American Enterprise Institute for Public Policy Research, 1977), 76 While Wertman nowhere specifically mentions illegality as a means of raising campaign funds, he repeatedly stresses that “little, in fact, is known about the sources of factional funds” (p. 77) 26 Della Porta and Vannucci (fn. 15), 236 21 v53.i4588golden 9/27/01 5:20 PM Page 597 Source: http://www.doksinet COMPETITIVE CORRUP TION 597 corruption, interparty competition may also augment it. If fear of seat loss drives attempts to

raise campaign fundsattempts that may easily slide into illegalityperhaps it does not matter whether the potential loss occurs because of rivalry within the same party or because of competition with candidates from other parties. To examine whether this is the case, we include measures of interparty competition in the multiple regressions analyzing the political causes of suspected malfeasance among DC members of parliament. Finally, we test the impact on the rise of corruption of the 1974 law regarding the public financing of Italy’s political parties. Whereas parties had previously been entirely self-financing, after 1974in what proved to be an ironic attempt to “clean up” campaign financingparties received public funding for their campaigns. At the same time, the law prohibited public companies from making contributions to parties and required parties to make annual public disclosure of their sources of funds.27 The law both criminalized much preexisting behavior and

eventuallywith the Clean Hands investigations of the 1990sprovided the legal grounds for prosecution of many politicians. MEASURES AND DATA THE DEPENDENT VARIABLE: POLITICAL CORRUPTION For obvious reasons, it is difficult to measure political corruption. By its nature, corrupt activity tends to be secret, and those involved have incentives to keep it that way. Studies of the phenomenon have relied on three different types of data in democratic contexts: expert or informed opinion, press reports, and/or legal proceedings or outcomes. An example of the firstindeed, probably the most prominent exampleis the comparative index of corruption made available by Transparency International on the basis of opinions of businesspeople.28 These data are useful for getting a broad-gauged idea of the extent of corruption in different countries Data limitations make it impossible to use this source for any substantial period of time or to analyze the causes of corruption within countries. Press

reports can also be utilized for assessing corruption, although, like all sources, their reliability is often unclear. They have been used in some empirical investigations.29 However, even where widespread cor27 For a description of the law in English, see Wertman (fn. 25), 77 Useful discussions include Thomas D. Lancaster and Gabriella R Montinola, “Toward a Methodology for the Comparative Study of Political Corruption,” Crime, Law and Social Change 27, nos. 3–4 (1997); and Johann G Lambsdorff, “Corruption in Comparative Perspective,” in Arvind K Jain, ed., Economics of Corruption (Boston: Kluwer, 1998) 29 For instance, Cazzola (fn. 14), chap 3 28 v53.i4588golden 9/27/01 5:20 PM Page 598 Source: http://www.doksinet 598 WORLD POLITICS ruption exists, the press may not expose it; or it may expose trivial incidents of graft but not have access to the information that would allow exposés of major networks of criminal behavior among politicians. Finally, gathering and

coding media reports is typically a laborious and analytically imprecise process. Court proceedings or outcomes, finally, can be used to measure corruption. Most of the existing empirical work on corruption in Italy uses legal records of various types, including the extremely useful studies by della Porta, della Porta and Vannucci, and Vannucci,30 which draw heavily on transcripts of the trials of those charged with political corruption. Our research strategy complements theirs In this study, we make use of a related legal source of information: charges of malfeasance against members of parliament. The reason for the frequency of such requests in postwar Italy lies with the unusually high guarantees of parliamentary immunity that protected parliamentarians. Until 1993 parliamentarians were protected from prosecution except when an absolute majority of the members of the house in which they served explicitly agreed to do otherwise.31 Not surprisingly, in the years after World War II,

most requests were not allowed to proceed and instead were denied or simply allowed to lapse for lack of response. One study finds that only 19 percent of RAP were approved during the years between 1948 and 1987, and in some legislatures (mainly in the early postwar era) the percentage was in the single digits.32 Regardless of the outcome, however, the data on the requests themselves are a useful source of information on suspected malfeasance by members of parliament. Parliament maintains its own records of all requests for authorization to proceed. These include the date of the request, the charges, the location of the magistrate’s office making the request (that is, the name of the city), the name of the MP being charged, and the eventual response by parliament (approval to proceed, rejection, or failure to act). Some research using these data has already been conducted.33 Cazzola generously provided his data set for use in the present analysis. It includes his coding of the house

in which the representative served 30 See della Porta (fn. 14); della Porta and Vannucci (fn 15); della Porta and Vannucci (fn 16); and Vannucci (fn. 15) 31 As of November 1993 the judiciary was automatically allowed to proceed unless an absolute majority voted against the request. This change in procedure greatly facilitated the Clean Hands investigations in the 1990s 32 Cazzola (fn. 14), 113, table IV1 33 Cazzola (fn. 14); Cazzola (fn 15); Cazzola and Morisi (fn 13); Franco Cazzola and Massimo Morisi, La mutua diffidenza: il reciproco controllo tra magistrati e politici nella prima Repubblica (Milan: Feltrinelli, 1996); and Ricolfi (fn. 15) v53.i4588golden 9/27/01 5:20 PM Page 599 Source: http://www.doksinet 599 COMPETITIVE CORRUP TION TABLE 1 CHARGES OF MALFEASANCE AGAINST DC MEMBERS OF THE TWO HOUSES OF PARLIAMENT, FIRST TEN LEGISLATURES a Legislature I (1948–53)b II (1953–58) III (1958–63) IV (1963–68) V (1968–72) VI (1972–76) VII (1976-79) VIII (1979–83) IX

(1983-87) X (1987–92)c Deputies Charged/Total (%) Senators Charged/Total (%) Difference % 34/306 (11) 24/262 (9) 35/273 (13) 28/260 (11) 22/265 (8) 51/265 (19) 26/262 (10) 38/262 (14) 46/226 (20) 42/234 (18) 12/150 (8) 5/111 (4) 10/121 (8) 6/134 (4) 7/137 (5) 16/136 (12) 13/136 (10) 20/139 (14) 11/121 (9) 13/127 (10) 3* 5* 5* 6* 3* 7* 0 0 11* 8* *significant at the 0.1 level or better; *significant at the 0.05 level or better; *significant at the 0.01 level or better SOURCES: Data courtesy of Franco Cazzola; originally provided by the Italian parliament. a Data include all charges of wrongdoing. b Results for the I Legislature are marginally significant. c Data for the X Legislature are incomplete. (Chamber or Senate), the MP’s partisan affiliation, and his or her electoral district, as well as the various categories of data that we have already enumerated as available in the original Italian parliamentary records. We supplemented this data set, which ends in 1987, with

data provided by a parliamentary office on charges of malfeasance against members of the Chamber of Deputies of the X and XI Legislatures. We coded the latter data for each MP’s electoral district. In using Cazzola’s coding distinguishing senators from deputies, we decided to drop the former from the analysis, both because senators were elected from single-member districts rather than by open-list PR and because we did not have data on charges of wrongdoing against members of the Italian Senate for the X and XI Legislatures. Our guiding hypothesis about the effects of open-list PR suggests that members of the Italian Senate should have been less exposed to the kinds of intraparty pressures that generated the need for illegal campaign funds, precisely because the Senate was not elected by open-list PR. To confirm this, Table 1 shows data comparing the proportion of DC deputies charged and the proportion of DC senators charged in each of the first ten postwar legislatures. (Note that

the proportion is measured as the total number of DC representatives charged divided by the total number of DC representatives elected to each chamber. The XI v53.i4588golden 9/27/01 5:20 PM Page 600 Source: http://www.doksinet 600 WORLD POLITICS Legislature is excluded because Cazzola’s data set extends only to 1987.) We tested whether the difference between the two proportions was significantly different from 0 for each legislature. The results reported in the table show that the proportion of DC deputies charged is significantly higher than the proportion of DC senators charged in eight out of ten legislative periods. As the hypothesis linking the electoral system to intraparty competition and then to malfeasance predicts, Christian Democratic senators tended to be less likely to be charged with malfeasance than were Christian Democratic deputies.34 We interpret the fact that the differences in the proportions (although statistically significant by conventional standards)

are often substantively not very large as indicating the contamination of all DC party politics by the incentives arising from the electoral system found in the Chamber of Deputies during this period. Senators were significantly less exposed to incentives to engage in political wrongdoing, but given the frequent crossover of politicians from one house to the other, they were far from immune. Readers will note from the data reported in Table 1 that a relatively small number of elected officials in the DC were investigated for any types of wrongdoing in the years between 1948 and 1987, when this data set ends. A wave of charges occurred in the period following 1987, and these subsequent charges (until the end of the XI Legislature in 1994) are included in the data set used for the remainder of this study. That data set is therefore considerably larger than the one in the analysis reported in Table 1. Two major issues affect the validity of our measure of political corruption. First,

requests to lift parliamentary immunity do not distinguish between allegations of corruption and allegations of other sorts of malfeasance. In principle, it should be possible to separate out charges of suspected corruption from other crimes, since the parliamentary records identify the legal statutes brought to bear in levying charges against the suspect.35 In practice, however, this is more difficult than might appear. Doing so requires interpreting the specific statutes cited. These run the gamut from murder to speeding Only one study of 34 We tested this hypothesis only for Christian Democratic legislators because our full data set did not separate out “opinion” crimes from other types of charges for members of other parties; for a discussion of types of charges, see pp. 600–601 35 Cazzola (fn. 14) distinguishes corruption from other types of suspected malfeasance, but he does not detail his coding procedures, making evaluation of them impossible. Hence, we chose not to use

his classification. His study classed 40 percent of the RAP against DC parliamentarians between 1948 and 1987 as involving political corruption. v53.i4588golden 9/27/01 5:20 PM Page 601 Source: http://www.doksinet COMPETITIVE CORRUP TION 601 which we are aware36 provides a classification scheme that distinguishes political corruption from other types of suspected malfeasance. However, because this classification has not been systematically evaluated in the literature, it is unclear whether it can legitimately serve as a standard. Furthermore, the high threshold for the suspension of parliamentary immunity in Italy introduces conceptual problems into any supposedly neat distinction between allegations of “corruption” and allegations of other types of wrongdoing. Precisely because it was so difficult to strip parliamentarians of their immunity and because parliamentarians were aware that they were likely to be protected from prosecution by their colleagues, almost any crime

could have involved some element of abuse of office. Unlike ordinary citizens, deputies and senators were largely protected from the legal consequences of breaking the law surely a subtle encouragement to wrongdoing. As a result, parliamentarians, knowing they would not be stripped of their parliamentary immunity for minor actsor even, in many cases, for major ones were little deterred from committing offenses. Nor is it surprising, therefore, that Italian commentators refer frequently to the extent to which the governing parties came to exhibit an “arrogance of power.” Any classification that interprets “corruption” simply as pocketing public monies or accepting illegal campaign contributions is likely to be excessively narrow in such a context. Our solution was to drop observations that we knew conclusively were unrelated to political corruption, namely, charges involving libel, slander, defamation, and other “opinion” crimes, including slander against the state and

slander of the public administration.37 This left in our data set all charges of wrongdoing conceivably related to abuse of public office and political corruption. The analysis presented in the remainder of this paper uses this reduced data set The second problem of validity is potentially more serious, in that we are less able to respond effectively to it. The measure of corruption that we use makes it impossible to distinguish suspected malfeasance on the part of legislators from the behavior of the judiciary, which is charged with investigating and prosecuting. While this problem is severe in principle, in practice it constitutes a contaminating measurement problem only if the judiciary’s behavior generates systematic bias in the RAP. If there is no systematic bias, then as in any statistical analysis, we simply assume that the variable is “noisy” As we detail below, we 36 Ricolfi (fn. 15), 151–64 This procedure was suggested to us by Davide Petrini. The exact statutes

dropped from the analysis were numbers 341–42 and numbers 594–99 37 v53.i4588golden 9/27/01 5:20 PM Page 602 Source: http://www.doksinet 602 WORLD POLITICS believe that of the various possible types of systematic bias that could characterize the data, only onepossible systematic underinvestigation of DC politicians between 1948 and the mid-1970sis likely to constitute a genuine problem. To anticipate, however, our statistically significant results do not bear on the period prior to the early 1970s, so the problem of potential data bias is simply not relevant to our findings. We now detail our full line of argument about possible biases in the data. The kinds of systematic biases we could observe include the systematic underinvestigation of alleged wrongdoing by deputies in the governing parties because of criminal or partisan associations on the part of the investigators; or systematic overinvestigation (that is, deliberate attempts to smear the reputations of deputies in

the governing parties) because of the partisan associations of the judiciary or factional conflicts among politicians themselves, leading some to denounce their rivals secretly to the judiciary. Systematic underinvestigation because of criminal associations among the judiciary appears to have been a relatively minor phenomenon. There are judges who are known associates of the Mafia, but this does not appear widespread, despite reports, such as that found in Gambetta,38 of instances in which judges repeatedly pronounced dubious rulings on Mafia suspects but were not removed or transferred. In any politically competitive situation, accusations of corruption may come, perhaps without adequate justification, from political opponents. Factions within the governing parties may systematically pass potentially incriminating information on to the judiciary in order to disable rivals from other factions in the same party. But whereas the political use of potentially incriminating information

seems relatively common in some other countries, the literature we have reviewed does not suggest that behavior along these lines occurred in Italy, where factional and partisan affiliation served instead to signal trustworthiness among the dishonest. Denunciations of others within the same party or among the ruling parties would have undermined the networks that had been established to bind the corrupt together, allowing predictable divisions of the spoils; moreover, such denunciations would have provided political ammunition to the main opposition party, the Italian Communist Party (PCI) , whose increasing ability to attract the votes of citizens dissatisfied with the system of DC rule came to pose a distinct political threat by the 1970s. The PCI’s relative lack of involvement in 38 Diego Gambetta, The Sicilian Mafia: The Business of Private Protection (Cambridge: Harvard University Press, 1993), 6–7. v53.i4588golden 9/27/01 5:20 PM Page 603 Source: http://www.doksinet

COMPETITIVE CORRUP TION 603 political corruption, its emphasis on “good government” and “the moral question,” and its growing electoral appeal made it politically all the more important that the communists (and the public generally) not be allowed access to incriminating information about members of the ruling parties.39 There was little incentive for members of the ruling cartel to turn to the judiciary with accusations against other members.40 There have been instances in which the judiciary has been known to issue requests to proceed for alleged “crimes” that were largely partisanpolitical in nature. In the early postwar era, not surprisingly, when most members of the judiciary were holdovers from the fascist era, numerous RAP were issued against communist MPs for alleged crimes involving “sabotage” against the Italian state and other purported efforts to undermine the constitution.41 Some have contendedquite plausibly that underinvestigation of political

corruption by members of the governing parties routinely occurred in the 1950s and 1960s.42 Likewise, others have argued43 that the RAP issued in the course of the Clean Hands investigations of the 1990s were largely politically motivated by an ideological judiciary consisting overwhelmingly of zealots of the left. Our reading of the evidence is that the facts do not substantiate the latter interpretation.44 Of possible instances of under- and overinvestigation of deputies due to judicial bias, only the systematic underinvestigation of the DC between 1948 and the mid-1970s is therefore empirically relevant. There is, given data now available, no way to know. If this occurred, then the results reported below for the first half of the legislatures that we ana39 The charge that by failing to denounce corruption in the ruling parties the PCI thereby colluded in it has been made by Silvia Colazingari and Susan Rose-Ackerman, “Corruption in a Paternalistic Democracy: Lessons from Italy for

Latin America,” Political Science Quarterly 113 (November 1998). It also underlies Alessandro Pizzorno’s characterization of Italy’s postwar political system as a “consociational” arrangement between government and opposition elites; see Pizzorno, “Le difficoltà del consociativismo,” in idem, Le radici della politica assoluta e altri saggi (Milan: Feltrinelli, 1993); and idem (fn. 15). This view fails to acknowledge the difficulties that must have been experienced by the PCI and such little independent press as existed in Italy in obtaining credible information about political corruption, given a system in which those involved in it systematically covered up for each other. 40 The same is not true of accusations of libel, slander, and defamation of character, which apparently often came from political rivals within the same party. By omitting cases involving such charges, we remove this potential source of bias 41 Since the present analysis is confined to the DC and

omits charges of insult of a public official (statute no. 341) and insult of the state (statute no 342), this does not pose a problem for us 42 Della Porta and Vannucci (fn. 16, 1999), 141–44 43 Burnett and Mantovani (fn. 16) 44 For instance, the magistrates involved in the Milan pool were themselves affiliated with a variety of political parties or with no party at all; some had left-wing political histories but some did not, and so on. Even Burnett and Mantovani (fn 16), although arguing this position, do not offer persuasive and thorough documentation supporting the contention that the Milanese judiciary was politically compact. v53.i4588golden 9/27/01 5:20 PM Page 604 Source: http://www.doksinet 604 WORLD POLITICS lyze may be biased. But if they are biased, it is in favor of the null hypothesiswhich, until the early 1970s, is precisely the result we report That is, we do not claim that systematic relationships exist between the independent variables we analyze and

charges of malfeasance against DC deputies in the years before the early 1970s. If we did, possible bias in the data would reduce our confidence in such a claim. Substantively, this means there may have been more political corruption in Italy in the 1950s and 1960s than we observe using these data, and indeed that analyses drawing on the RAP are unlikely to generate systematic statistical results for this period. It may be difficult ever to know the causes of political corruption in Italy over the entire postwar era, given that for much of the period, so much presumed corruption remained uninvestigated. But this in no way invalidates the results of our analysis for the period starting in 1972.45 THE INDEPENDENT VARIABLES: INTRAPARTY AND INTERPARTY COMPETITION In this section we discuss our measure of intraparty competition and then the ways we operationalize interparty competition. As we have already argued, Italy’s open-list electoral system forced candidates of the same party to

compete with each other. As recent work on Brazil has noted, even in electoral systems that encourage “the personal vote,” specific political parties may overcome the incentives toward personalism, especially if they are without access to the resources offered by control of government itself.46 This is precisely the case for the Italian Communist Party, a well-disciplined and centralized organization denied national power throughout the period. For reasons that lie beyond the scope of the present article, the party successfully overcame the incentives of the electoral system toward intraparty competition and factionalism, instructing its electors in the distribution of their preference votes47 where it did not simply discourage preference voting altogether. Hence, in the 1970s, for instance, the PCI received the smallest number of preference votes of any of Italy’s five largest parties.48 The Christian Democrats, by contrast, found themselves dividing 45 Ultimately, only

additional research into the postwar Italian judiciary will allow us to distinguish more fully our measure of suspected political wrongdoing from judicial independence. One possible measure of potential judiciary autonomy is the extent of resources available to different judicial offices throughout the country over time, but formidable data collection problems would have to be confronted. 46 Samuels (fn. 5) 47 Wertman (fn. 25), 75 48 See the estimates in Richard S. Katz, “Preference Voting in Italy: Votes of Opinion, Belonging, or Exchange,” Comparative Political Studies 18 ( July 1985), 233, table 1. v53.i4588golden 9/27/01 5:20 PM Page 605 Source: http://www.doksinet COMPETITIVE CORRUP TION 605 into factions by the mid-1950s, as politicians competed for internal resources and sought to build patronage machines capable of lining up preference votes.49 By the 1960s the DC’s factions were formally organized, each with its own headquarters, news agency, regularly scheduled

meetings, and lists for party office.50 The party adopted PR for internal party elections beginning in 1971,51 so that factions enjoyed formal representation in DC governing bodies according to their relative strength. By the 1980s factional strength on the DC’s National Council was calculated down to hundredths of percentages in the allocation of cabinet positions.52 The National Council was in turn elected from within the party, and competition for both party and government positions was regulated largely by counting “the amount of preference votes obtained in party and government elections.”53 Amassing preference votes was hence critical not only to winning office but also to controlling resources within the national party organization. DC factions were generally not ideologically distinct (although they can be arrayed ideologically) but instead were largely organized around the patronage machines controlled by individual politicians, mainly national parliamentarians. These

factions constituted the cartel that governed Italy from 1948 until the so-called opening to the left in 1963, when the PSI first entered a coalition government with the DC. This enlarged the cartel to include the Socialists, most notably; minor cartel members during the postwar era also included the Italian Liberal Party (PLI), the Italian Republican Party (PRI), and the Italian Social Democratic Party (PSDI). There is now considerable documentation that all governing parties were systematically involved in kickbacks and corruption, that agreements existed among them regulating the proportion of the spoils each would receive, that corrupt exchanges involved large numbers of national politicians, and that they covered up for each other (in part by voting against lifting parliamentary immunity of those charged with corruption). The complex networks across parties, between nationally organized factions and local political machines, from electoral officials to bureaucrats in the public

administrationmembers 49 Gianfranco Pasquino, “Le radici del frazionismo e il voto di preferenza,” Rivista Italiana di Scienza Politica 2 (August 1972). 50 Alan S. Zuckerman, The Politics of Faction: Christian Democratic Rule in Italy (New Haven: Yale University Press, 1979), 111. 51 Giovanni Sartori, “Proporzionalismo, frazionismo e crisi dei partiti,” Rivista Italiana di Scienza Politica 1 (December 1971). 52 Reported by Carol Mershon, The Costs of Competition (Stanford, Calif.: Stanford University Press, forthcoming.) 53 Zuckerman (fn. 50), 67 v53.i4588golden 9/27/01 5:20 PM Page 606 Source: http://www.doksinet 606 WORLD POLITICS of each of them involved in a gigantic web of corruption, kickbacks, and bribesis beyond the scope of the present analysis. This said, it is important to note that although local politicians apparently were to some extent free to determine the division of the bribes that would occur locally, the Clean Hands investigations revealed a system

of corrupt exchanges in which national party leaders the members of parliament studied herewere active participants and possibly the initial instigators. We study members of parliament mainly because of the availability of systematic data, but this choice is justified by the fact that the trail of evidence uncovered in the investigations that took place in the 1990s clearly led to them, revealing a system of corruption that was ultimately national in scope. Political corruption in postwar Italy involved many persons beyond those studied here, but Christian Democratic deputies were critical figures in the networks of corrupt exchanges that developed. Although measures of intraparty conflict in the DC at the level of electoral districts are not readily available, a good proxy can be constructed. This consists of summing the number of preference votes given to DC candidates for the Chamber of Deputies in each electoral district and deflating the sum by the total number of list votes

received by the party in the same election and same district, for each of the country’s thirty-one (until the 1958 election, thirty) electoral districts that used PR.54 Until the 1992 elections Italy had an extremely pure system of PR.55 (In 1994 Italy adopted a mixed system of PR and majoritarianism, in which the latter predominated.) Under the PR system, the parties listed candidates for each electoral district, each of which sent an average of twenty representatives to the Chamber. Voters marked their ballots with their party of choice (known as list votes). At the same time, voters were allowed to indicate up to four “preferences” for individual candidates listed by the party for which they voted, although they could choose to exercise fewer or even no preferences.56 Only about a third of all possible preference votes were normally cast.57 A 1991 referendum 54 Italy’s smallest region, the Valle d’Aosta, was, because of its size, a single-member district that used a

plurality system. Because intraparty competition did not occur there, the district is not included in the analysis reported below. 55 For descriptions in English, see Samuel H. Barnes, Representation in Italy: Institutionalized Tradition and Electoral Choice (Chicago: University of Chicago Press, 1977), chap 3; and Wertman (fn 25) 56 Districts electing sixteen or more representatives allowed a maximum of four preference votes to each elector; those electing fewer allowed three; reported in Barnes (fn. 55), 36 57 Richard S. Katz and Lucio Bardi, “Preference Voting and Turnover in Italian Parliamentary Elections,” American Journal of Political Science 24 (February 1980), 99 v53.i4588golden 9/27/01 5:20 PM Page 607 Source: http://www.doksinet COMPETITIVE CORRUP TION 607 resulted in the reduction of preference votes to only one,58 so the analysis of the 1992 parliamentary elections reported below uses the new measure. The actual individuals seated in parliament were those who

had received the most preference votes. Getting a large number of preference votes was thus essential to attaining or retaining officeindeed, within the DC, the single most important cause of turnover in the two parliamentary elections held in the 1970s was intraparty defeat59and also important for acquiring or retaining influence within the party. We restrict the present analysis to the DC, for three reasons. First, the party was factionalized for a much longer period than other Italian parties, thereby permitting a longer valid time period for analysis. Second, familiarity with Italian history suggests that postwar political corruption originated in the DC Third, we currently have data on preference votes for DC candidates only.60 In principle, the analysis undertaken here could be extended to include all the governing parties, and it would be especially interesting to enlarge the analysis to include the Italian Socialist Party. The data on preference votes received by individual

candidates, as well as district-level electoral data, are available for every postwar election in a series of volumes commonly known as La Navicella, after the publisher.61 We constructed a data set that included the name of each DC candidate for each district in each election, the number of preference votes received by each, and whether the candidate was subsequently seated in the Chamber.62 We also included the number of list votes received by the DC in each election and each district,63 as well as the elec58 For an analysis, see Gianfranco Pasquino, “La preferenza unica: reazioni e opportunità,” in Gianfranco Pasquino, ed., Votare un solo candidato: le consequenze politiche della preferenza unica (Bologna: Il Mulino, 1993). 59 Katz and Bardi (fn. 57) 60 Figures on preference votes have to be assembled and electronically input by hand, candidate by candidate, district by district, and election by election. 61 The volumes are available starting with La Navicella, I Deputati e

Senatari del Prima Parlamento Repubblicano (Rome: La Navicella, 1948). 62 There were exceptions to the rule that candidates receiving the most preference votes received parliamentary seats. Occasionally, a candidate died before taking his seat More common was the phenomenon of standing in multiple districts, since Italian law permitted parliamentary candidates to stand in as many as three districts, as well as to run for both the Senate and the Chamber simultaneously. In these cases, candidates who received enough preference votes in different districts to win more than one parliamentary seat were required to select the district they would represent as well as the house in which they would be seated. Hence, there were instances in which a candidate who received enough preference votes to win a Chamber seat did not in fact subsequently take that seat, instead opting for another. 63 Where necessary, we supplemented the information available in La Navicella with data from Gianfranco

Pasquino, ed., 1945–1996: archivio della politica in Italia (Rome: Laterza Multimedia, CD Rom, 1996). v53.i4588golden 9/27/01 5:20 PM Page 608 Source: http://www.doksinet 608 WORLD POLITICS toral quota required in each election and each district to win a Chamber seat and the total number of deputies elected from each district. It seems plausible that the increases in political competition and the seat loss that affected DC members of parliament with the rise of the left over the postwar era may have fueled the growth of corruption among Christian Democrats, as MPs sought to raise funds to protect themselves from losing office. In 1948 the DC controlled 485 percent of seats in the Chamber of Deputies; in 1963 this fell drastically to 38.3 percent. Another sharp decline occurred in 1983, when the DC’s share fell to 32.9 percent Thus over time both the party and many individual DC politicians came to be increasingly exposed to interparty competition. We constructed three

separate measures of interparty competition: (1) the change in the proportion of list votes received by the DC in each election, (2) the change in the proportion of seats held by the DC in each election, and, finally, (3) a measure of marginality, constructed as the ratio between the number of votes received by the lowest-ranked DC candidate who won a seat (that is, the winning DC candidate who garnered the smallest number of preference votes among party contenders) and the electoral quota for the district (that is, the number of votes required to win a seat). The last measure captures how “safe” the most marginal DC seat is.64 We constructed each of these three measures using three different types of lags: the change at time t over t–1; the change at time t–1 over t–2; and the change at time t over t+1. The reason for using different lags was that, although we believed that illegal activities among political candidates would be greatest in the period just prior to national

elections, we had no theory of when candidates would feel most threatened by other parties. Various scenarios seemed plausible: perhaps vote or seat loss in one election would engender more illegal fund-raising in the next period; perhaps the threat of vote or seat loss would engender contemporaneous illegal fund-raising; or perhaps, finally, illegal fund-raising occurred in anticipation of electoral losses. ANALYSIS AND RESULTS To examine whether intra- and interparty conflict were significant determinants of suspected political corruption of Christian Democratic legislators, we set up a series of regressions of the following form: 64 As far as we know, there is no standard measure of marginality in PR settings. This one seemed to us a reasonable approximation. Alternatives that occurred to ussuch as the difference between the number of preference votes received by the highest- and second-highest-ranking DC candidatescaptured intraparty competition more than interparty competition.

v53.i4588golden 9/27/01 5:20 PM Page 609 Source: http://www.doksinet 609 COMPETITIVE CORRUP TION SUSCOR = β0it + β1it INTRA + β2it INTER + Σ γ Year + ε 10 t it (1) t=1 where i is the electoral district t is the year of the legislative election SUSCOR is suspected corruption on the part of DC deputies, measured as the number of nonopinion crime RAP lodged against DC MPs over the life of the subsequent legislature divided by the number of DC MPs (or the proportion of DC deputies charged)65 INTRA is intraparty competition, measured as the total number of preference votes received by Christian Democratic candidates divided by the total number of list votes received by that party in the same district INTER is interparty competition, measured as the share of DC votes (model 1) or the share of DC seats (model 2) or the number of preference votes received by the winning DC candidate with the lowest number of preference votes over the electoral quota (model 3)66 YEAR is

the election year dummy variable to pick up any variance unaccounted for by the main predictors and ε is an error term Parliamentary elections were held in 1948, 1953, 1958, 1963, 1968, 1972, 1976, 1979, 1983, 1987, and 1992. Table 2 reports the results of pooled regressions that were run using each of the three measures of interparty competition detailed above. We report panel-corrected standard errors in order to guard against potential problems of heteroskedasticity across districts and contemporaneous correlation of errors across years.67 The table shows the results only for those regressions using t – t–1, but results with alternate lags were similar68 First, notice in Table 2 that the probability of a large-F distribution is low enough to allow one to reject the null hypothesis that there is no relationship between the explanatory variables and the dependent variable at the 0.001 level in all three models These models all reveal strong and significant relationships between

our independent and dependent variables. Even more striking is that the coefficients for INTRA are positive and significant at the 0.01 level in all three models, whereas 65 Since multiple sets of charges may be lodged against the same individual deputy, where a single was charged more than once during the life of a single legislature, we dropped the multiple observations, thereby capping the number of RAP against a single MP at 1 for any legislature. 66 Each measure of the variable INTER was constructed using each of the three lags discussed above, generating nine separate regressions. 67 Nathaniel Beck and Jonathan N. Katz, “What To Do (and Not To Do) with Time-Series CrossSection Data,” American Political Science Review 89 (September 1995) 68 Complete results are available from the authors upon request. MP v53.i4588golden 9/27/01 5:20 PM Page 610 Source: http://www.doksinet 610 WORLD POLITICS TABLE 2 RESULTS OF POOLED REGRESSIONS TESTING FOR THE IMPACT OF INTER- AND

INTRAPARTY COMPETITION ON CHARGES OF DC MALFEASANCE, FIRST ELEVEN POSTWAR LEGISLATURES Independent Variable INTRA INTER N Adjusted r 2 Prob.> chi 2 Model 1 (Vote Share) Model 2 (Seat Share) Model 3 (Marginality) 0.0482* (0.0229) 0.3501 (0.4185) 309 0.5114 0.0000 0.0503* (0.0234) –0.0325 (0.0675) 309 0.5100 0.0000 0.0404* (0.0249) 0.0743 (0.0319) 309 0.5161 0.0000 *significant at the 0.1 level or better; *significant at the 0.05 level or better; *significant at the 0.01 level or better; panel-corrected standard errors in parentheses the coefficients for INTER are not. This result obtains in regressions using all three ways of lagging the variable INTER, although the table reports only one.69 Finally, note that our two political variables alone account for about half of the variation in the dependent variable in each of the three models, underscoring the importance of political explanations of high-level corruption. Because the units of measurement are not intuitively

meaningful, we cannot interpret the size of the effect of intraparty competition on alleged malfeasance. Our results indicate that charges of malfeasance against DC Deputies for nonopinion crimes tend to be positively affected by intraparty competition and that they are uncorrelated with interparty competition. The absence of significant results for any of the measures of interparty competition that we used corroborates the standard interpretation of the Italian electorate as largely frozen in its electoral orientations.70 Competition with other parties was generally so limited that it did not systematically and significantly affect the likelihood of DC candidates engaging in illegal fund-raising activities. Instead, it seems likely that competition with other parties was mainly handled with resources distributed by the DC ’s national party headquarters. Competition with 69 To facilitate presentation, coefficient estimates for election-year dummy variables are not included; they are

significant for every legislature starting in 1972 except 1976. 70 Arturo Parisi and Gianfranco Pasquino, eds., Continuità e mutamento elettorale in Italia: le elezioni del 20 giugno 1976 e il sistema politico italiano (Bologna: Il Mulino, 1977); in English, see Arturo Parisi and Gianfranco Pasquino, “Changes in Italian Electoral Behavior: The Relationship between Parties and Voters,” in Peter Lange and Sidney Tarrow, eds., Italy in Transition: Conflict and Consensus (London: Frank Cass, 1980) v53.i4588golden 9/27/01 5:20 PM Page 611 Source: http://www.doksinet COMPETITIVE CORRUP TION 611 other candidates from within their own party drove the allegedly illegal behavior exhibited by Christian Democratic deputies. To examine possible temporal variation across legislatures, Table 3 reports the results of separate regressions for each of the eleven legislatures.71 The results in Table 3 are statistically significant by conventional standards for our measure of intraparty

competition for four of the six legislatures starting with the VI, that is, as of the 1972 legislative elections. As of this date DC intraparty competition significantly affected charges of malfeasance against DC deputies Interparty competition, by contrast, is significant only in 1976when it had the effect opposite of the one predicted, lowering the number of incidents of suspected malfeasanceand again in the final legislative elections studied, when an extremely large proportion of deputies were subject to investigation (substantially increasing the number of cases included in the analysis). Like the earlier pooled analysis, these results also corroborate the overriding importance of intraparty competition in affecting charges of malfeasance against DC deputies and the relative unimportance of interparty rivalry. The 1974 change of law regarding the public financing of Italy’s political parties makes us suspect this may have contributed to the fact that the results are

statistically significant only starting with elections to the VI Legislature. To investigate this, we performed two Chow tests,72 looking for structural breaks in the data, one where we broke the data as of 1972 and a second where we broke it as of 1976. The resulting F values enable us to reject the null hypothesis that there is no structural break at the 0.001 level for both tests This corroborates the impression that the data are characterized by an important change as of the early or mid-1970s, but we cannot tell if the break occurred just prior to the 1974 lawwhich was passed in the middle of the VI Legislatureor just after. Given the coincidence in timing, however, we suspect that political corruption in Italy was reinforced by the 1974 change in the law regulating the public financing of political parties. One interpretation is that the law allowed the judiciary to proceed with investigations of politicians suspected of wrongdoing more aggressively 71 In the results reported

interparty competition is operationalized as the loss of the share of DC seats in the district at time t over t–1 (model 2). The result for the first postwar election is therefore omitted, since there was no preceding election. While the other two models yield similar results, they are not presented to facilitate exposition. Complete results are available from the authors upon request 72 Briefly stated, the Chow test works as follows. First, we set up a dummy variable, PERIOD, which equals 1 after 1972 (or 1976) and 0 otherwise. Then we create two interaction variables, PXINTER and PXINTRA, by taking the product of PERIOD and INTER and the product of PERIOD and INTRA. Finally, we examine whether the coefficients of PERIOD, PXINTER, and PXINTRA are jointly different from 0. N/A N/A N/A N/A INTRA INTER INTERCEPT N r2 –.009 (.026) .159 (.110) .067 (.048) 30 .035 Leg. II 1953–58 .026 (.031) .135 (.125) .046 (.051) 31 .043 Leg. III 1958–63 –.012 (.034) –.089 (.154)

.091* (.055) 31 .016 Leg. IV 1963–68 –.001 (.024) –.052 (.110) .062* (.035) 31 .007 Leg. V 1968–72 .102* (.047) .228 (.250) .018 (.062) 31 .174 Leg. VI 1972–76 .052 (.036) –.313* (.144) .019 (.046) 31 .205 .063* (.040) –.233 (.197) .038 (.056) 31 .141 Leg. VII Leg. VIII 1976–79 1979–1983 .146* .012 (.065) (.041) –.424 .070 (.355) (.205) .000 .137* (.073) (.072) 31 31 .246 .004 .850* (.269) .428* (.274) –.198 (.179) 31 .270 Leg. IX Leg. X Leg. XI 1983–87 1987–92 1992–94 5:20 PM *significant at the 0.1 level or better; *significant at the 0.05 level or better; *significant at the 0.01 level or better a Robust standard errors are in parentheses. b N/A = not available because no prior election. Leg. I 1948–53b 9/27/01 Variable TABLE 3 RESULTS OF REGRESSIONS TESTING FOR THE IMPACT OF INTER- AND INTRAPARTY COMPETITION ON CHARGES OF DC MALFEASANCE, FIRST ELEVEN POSTWAR LEGISLATURESa (MODEL 2, SEAT SHARE) v53.i4588golden Page 612 Source:

http://www.doksinet v53.i4588golden 9/27/01 5:20 PM Page 613 Source: http://www.doksinet COMPETITIVE CORRUP TION 613 than before, when legal grounds for prosecution were weaker, even in cases of accepting kickbacks. But it is also possible that the law “created” political corruption by the simple fact of criminalizing previously licit behavior. II. INTERPRETING THE RESULTS: CULTURAL DIFFUSION VERSUS LEGAL CHANGE Why are charges of malfeasance against Christian Democratic deputies significantly associated with intraparty competition only as of the early 1970s and not earlier? This finding can be interpreted in at least two distinct ways. The first is institutional: the 1974 change in campaign financing laws criminalized behavior that had previously been legal Systematic changes in behavior among DC politicians did not necessarily occur over the period studied. Supporting this hypothesis is the fact that in the 1990s, when the investigations of a majority of members of the

Italian parliament occurred, a vast number of them were charged with crimes pertaining to the violation of the law on party financing. Results presented above suggest that the data display a structural break about 1974. An alternative interpretation is evolutionary. According to this view, factionalism generated corruption early in the postwar era, but the relationship was weak and unsystematic. Eventually, as increasingly large numbers of Christian Democratic deputies used ever larger sums of illicitly gained campaign funds, the relationship became systematic. By the mid-1970s it was substantial enough to generate a statistically significant coefficient in regression analysis. Arguably, the latter is the conventional interpretation found in the Italian literature, according to which the illegal use of funds for political campaigns begins on a piecemeal and occasional basis. Over time assuming that nothing, such as expulsion from public office, interrupts the growth of such a

systemselection mechanisms make corruption in campaign financing the norm rather than the exception. Alessandro Pizzorno has argued that political corruption may become endemic when the funds gained provide a sufficient advantage to the corrupt such that they enjoy a higher probability of winning each electoral encounter.73 In such a situation, politicians who are not prepared to adopt such strategies are wiped out (that is, they lose their seats), while those who are not initially corrupt but manage to survive electorally are 73 Pizzorno (fn. 15) v53.i4588golden 9/27/01 5:20 PM Page 614 Source: http://www.doksinet 614 WORLD POLITICS forced to lose whatever scruples they may have held and imitate their more successful colleagues. Eventually we intend to study this question directly by examining the career paths of DC politicians charged with corruption. Data limitations currently prevent this74 For the moment, we examine a specifically cultural variant of the evolutionary

argument The logic of this is well known, thanks to the work of Robert Putnam.75 As Putnam, like others before him,76 has argued, southern Italy exhibits distinct patterns of political culture, patterns marked by high rates of preference voting, low turnout in referenda, low newspaper readership, and a scarcity of sports and cultural associations77as well as, presumably, higher rates of organized crime and political corruption. The cultural perspective can be easily adapted to the phenomenon of political corruption. The standard story is that corruption spread in Italy from south to north along with the clientelistic political machine of the DC. Hine, for instance, argues this quite explicitly, as well as offering a plausible explanation for the initial emergence of corruption in the Italian south: Any region starved of resources, but operating in a highly centralised state, with a uniform tax and welfare system, will demand of its representatives that they spend a major part of their

time competing for resources for their localities. If the region in question has . a high level of organised crime with perpetrators anxious to use their resources to infiltrate the political and financial systems to recycle illegal earnings and buy political protection . that too will ultimately have an effect on national political life.78 And he sums up the consequences by claiming that the political system in the north “seems to have acquired many of the characteristics originally most prevalent in the south.”79 While the same study later examines institutional factors responsible for the widespread political corruption observed throughout Italy, including the preference vote, the geographic contagion argument is widespread in the literature. It holds that corruption is not endogenous to institutions of the Italian political system but rather was introduced 74 Because of difficulties in matching the records by name, we have not been able to combine our data sets listing

the legislators charged with malfeasance with the data sets we compiled reporting the number of preference votes DC candidates received. We expect to do this in the future 75 Robert Putnam, Making Democracy Work: Civic Traditions in Modern Italy (Princeton: Princeton University Press, 1993). 76 Most notably, Edward C. Banfield, The Moral Basis of a Backward Society (Glencoe, Ill: Free Press,1958). 77 Putnam (fn. 75), 96 78 Hine (fn. 16), 145 79 lbid. v53.i4588golden 9/27/01 5:20 PM Page 615 Source: http://www.doksinet COMPETITIVE CORRUP TION 615 from outside of it, thanks to historical and cultural traits of southern Italy, including those involving the Mafia and organized crime.80 These factors interacted with the long-standing economic gap between northern and southern Italy to drag southern politicians into corrupt activities, which then spread to the rest of the country via a process of cultural contagion. To examine this interpretation, we map intraparty competitiveness

and suspected corruption over the course of the postwar era. Figures 2–5 show both intraparty DC competition and charges of corruption against Christian Democratic MPs by electoral district for the two earliest postwar legislatures and the two final ones under the old electoral system.81 The values of each variable were pooled across Italy’s first eleven postwar legislatures so that the shading on the maps is comparable across legislatures. The pooled data were quartiled, and each quartile was shaded differently, representing least to most intraparty competitiveness/corruption. The only exception to this standardization procedure was the measure of factionalism for the 1992 election, which allowed only a single preference vote to electors. As pooling these results with those from other elections would artificially reduce the variation across electoral districts in 1992, we did not do so Examination of the maps showing intraparty competition confirms the conventional wisdom that the

use of preference votes has always been more extensive in southern Italy than in northern Italy.82 Intraparty competitiveness appears relatively limited in the first postwar election. The maps corroborate other studies that have argued that intraparty competition and factional conflict in the DC underwent an initial increase in the early to mid-1950s83 There is no substantial evidence of a nationalization of intraparty competitiveness in the DC even in the late 1980s or early 1990s. Preference voting remains a largely southern activity. The maps showing the proportion of DC members of parliament charged with nonopinion types of malfeasance in the early and later postwar legislatures are quite different from what the cultural hypothesis predicts. First, politicians charged with wrongdoing were not con80 This view turns on classing the Mafia as a cultural phenomenon, not an institution. The intervening maps are not presented for reasons of space but are available from the authors upon

request. 82 See Franco Cazzola, “Partiti, correnti e voto di preferenza,” in Mario Caciagli and Alberto Spreafico, eds., Un sistema politico alla prova: studi sulle elezioni politiche italiane del 1972 (Bologna: Il Mulino, 1975), 143; and Giovanni Schepis, “Analisi statistica dei risultati,” in Alberto Spreafico and Joseph La Palombara, eds., Elezioni e comportamento politico in Italia (Milan: Edizioni di Comunità, 1963), 376. 83 Zuckerman (fn. 50), 110–11 81 v53.i4588golden 9/27/01 5:20 PM Page 616 Source: http://www.doksinet DC INTRAPARTY FIGURE 2 COMPETITIVENESS AND CHARGES OF MALFEASANCE AGAINST DC MP’S: I LEGISLATURE, 1948–53 v53.i4588golden 9/27/01 5:20 PM Page 617 Source: http://www.doksinet FIGURE 3 CHARGES OF MALFEASANCE AGAINST DC MP’S: II LEGISLATURE, 1953–58 DC INTRAPARTY COMPETITIVENESS AND v53.i4588golden 9/27/01 5:20 PM Page 618 Source: http://www.doksinet DC INTRAPARTY FIGURE 4 COMPETITIVENESS AND CHARGES OF MALFEASANCE AGAINST

DC MP’S: X LEGISLATURE, 1987–92 v53.i4588golden 9/27/01 5:20 PM Page 619 Source: http://www.doksinet DC INTRAPARTY FIGURE 5 COMPETITIVENESS AND CHARGES OF MALFEASANCE AGAINST DC MP’S: XI LEGISLATURE, 1992–94 v53.i4588golden 9/27/01 5:20 PM Page 620 Source: http://www.doksinet 620 WORLD POLITICS centrated in the south in the early postwar years. The first two legislatures saw only a single southern district in the highest quartile of those charged.84 Second, as charges of wrongdoing became more frequent later in the postwar era, they were also distributed across all parts of Italy. (In the VIII Legislature, for instance, nine of Italy’s thirty-two electoral districts appear in the highest quartile of proportion of MPs charged, double the number of districts so classed in the I Legislature.) Suspected wrongdoing appears at high levels in districts from the top of the peninsula to the bottom of the boot. By the time the Clean Hands investigations were under way,

political corruption in Italy had become a national phenomenon. In the final legislature we study, twenty-three districts appear in the highest quartile, and eleven of these are located in the south. It is possible, of course, that the measure employed here has produced a geographically misleading picture of suspected criminal behavior among Christian Democratic politicians. The southern judiciary may have been less prone than the northern judicial offices to investigate suspected wrongdoing by DC deputies, perhaps because organized crime often figured as an intermediary in political corruption in the south. We have no way of verifying whether data on the RAP are biased in this way, although it seems probable. But if they are, then the relatively high incidence of suspected wrongdoing among northern DC deputies who served in the first and second postwar legislatures makes us suspect that were more valid data available, they would only reinforce the finding that suspected political

corruption has always been dispersed across the Italian electoral districts and was not initially concentrated in the south. The reason is that if the present data set is biased, this bias means that we are undercounting suspected corruption in the south in the early legislatures. A less biased data set would therefore show suspected corruption in the earlier postwar era not only in the north but also in the south (unless the underreporting of wrongdoing in the south is so massive in the early period as to dwarf even the relatively high levels of charges brought against deputies in some northern districts). The requests to remove parliamentary immunity of DC deputies suggest that political corruption, rather than emerging out of Italy’s supposedly familial southern political culture as Banfield and Putnam would contend, was instead a product of its postwar political institu84 In the III Legislature four southern districts were classed in the highest quartile of suspected malfeasance.

v53.i4588golden 9/27/01 5:20 PM Page 621 Source: http://www.doksinet COMPETITIVE CORRUP TION 621 tions. It was manifest at high levels in various electoral districts across Italy from relatively shortly after World War II. III. CONCLUSIONS We have demonstrated that Italy’s use of the preference vote appears to have contributed significantly to the development of extensive, highlevel political corruption in the postwar era. This relationship is apparent as of the early 1970s, perhaps in connection with the change in a law regulating party financing. We interpret these findings as demonstrating that political corruption emerged as a result of the search for campaign funds by DC members of parliament in their competition with candidates from the same party. We find no evidence in our data that corruption was initially significantly higher in southern Italy or that it spread from south to north. Our finding that interparty competition typically made no statistically significant

contribution to political corruption should not be interpreted as meaning that characteristics of the Italian party system were not relevant to the growth of corruption there. Underlying the growth of corruption among DC deputies was the justifiable belief that the PCI did not constitute a genuine threat to the DC’s hold on national office85that the combination of political veto by the United States to the entry of a Communist Party into a NATO government and anticommunism on the part of much of the Italian electorate would continue to allow the DC to remain permanently in office. Indirect corroboration of this interpretation comes from the fact that only after the fall of the Berlin Wall, the subsequent demise of the U.S veto on PCI participation in government, and the related shifts in electoral behavior was the system of political corruption exposed and then brought down.86 Until then, the DC and its allies could continue to believe that engaging in extensive illegal activities

would carry no major political consequences. This “arrogance of power,” as it is commonly called in Italy, pushed Italy well off the regression line in the comparative extent of corruption. 85 Sidney Tarrow, “Maintaining Hegemony in Italy: ‘The Softer They Rise, the Slower They Fall!’” in T. J Pempel, ed, Uncommon Democracies: The One-Party Dominant Regimes (Ithaca, NY: Cornell University Press, 1990). 86 We do not believe, however, that the end of communism was a significant cause in exposing Italian political corruption or in the collapse of the postwar party system; for a discussion, see Miriam A. Golden, “International Sources of the Collapse of Rent-Seeking Regimes: Hypotheses Drawn from the Italian Case” (Paper presented at the Joint Sessions of the European Consortium for Political Research, Grenoble, France, April 6–11, 2001). v53.i4588golden 9/27/01 5:20 PM Page 622 Source: http://www.doksinet 622 WORLD POLITICS To return to our opening puzzles,

because of the absence of political alternation, competition of elites in Italy was abnormally restricted. This prevented a proper functioning of legal and constitutional protections against political corruption, magnifying the pernicious effects of specific aspects of the electoral system. These institutional aberrations permitted excessive and persistent misgovernment by elected officials. The collapse in the 1990s of the international and domestic conditions that bred such a poorly functioning democratic political order bode well for a substantial, and possibly permanent, reduction of political corruption in Italy.87 Our analysis demonstrates the importance of political institutions in significantly affecting the degree of political corruption and, by extension, the extent to which deliberate and appropriate institutional modifications may reduce the incentives for corrupt behavior by legislators. 87 Given this, readers may wonder why the TI index for Italy did not show a

substantial decline even at the turn of the millennium. The reason is that the TI index fails to distinguish bureaucratic from political corruption In fact, since it is based on surveys of businesspeople, it may even reflect mainly bureaucratic corruption We believe that bureaucratic corruption has remained relatively high in Italy (in part due to the absence of thoroughgoing reform of the public administration), whereas political corruption has fallen drastically since the change of regime in the mid-1990s. On the persistence of bureaucratic corruption, see Alberto Vannucci, “Inefficienza amministrative e corruzione,” Rivista Trimestrale di Scienza della Amministrazione 44, no. 1 (1997)